發(fā)布時(shí)間：2018-06-21 08:04

  本文選題：情感計(jì)算 + 圖像情感��； 參考：《哈爾濱工業(yè)大學(xué)》2016年博士論文

【摘要】：隨著計(jì)算機(jī)科學(xué)、多媒體技術(shù)以及社交網(wǎng)絡(luò)的迅速發(fā)展,圖像、視頻等多媒體內(nèi)容的規(guī)模呈指數(shù)式爆炸增長(zhǎng),處理和理解這些多媒體內(nèi)容的需求日益增強(qiáng)。相對(duì)于底層視覺(jué)特征層,人們只能夠感知和理解圖像、視頻的高層語(yǔ)義層,包括認(rèn)知層和情感層。以往對(duì)圖像內(nèi)容分析的工作主要集中在理解圖像的認(rèn)知層,即描述圖像的真實(shí)內(nèi)容,如物體檢測(cè)與識(shí)別。然而,公眾對(duì)數(shù)字?jǐn)z影技術(shù)的廣泛使用及對(duì)圖像情感表達(dá)的強(qiáng)烈需求,使得對(duì)圖像最高語(yǔ)義層—情感層的分析變得越來(lái)越迫切。對(duì)圖像情感層的分析,簡(jiǎn)稱圖像情感計(jì)算,主要目的是理解觀察者看完圖像后所引起的情感反應(yīng)。圖像情感計(jì)算的發(fā)展主要受到兩大挑戰(zhàn)的制約：一個(gè)是情感鴻溝,即“可度量的信號(hào)屬性即特征與人感知該信號(hào)所期望產(chǎn)生的情感之間的不一致性”；另一個(gè)是人類情感感知與評(píng)估的主觀性,即“由于文化背景、教育程度、社交上下文等多種因素的影響,不同觀察者對(duì)同一幅圖像的情感感知是主觀的、不同的”。本文針對(duì)圖像情感計(jì)算中的上述問(wèn)題進(jìn)行研究,基于藝術(shù)學(xué)相關(guān)理論,期望提取更具有判別力更容易理解的情感特征；利用社交媒體數(shù)據(jù)進(jìn)行以用戶為中心的個(gè)性化情感預(yù)測(cè),探索社交媒體中影響情感感知的因素；對(duì)圖像情感的分布進(jìn)行建模,預(yù)測(cè)一幅圖像在多位觀察者中所誘發(fā)情感的分布情況；研究圖像情感在計(jì)算機(jī)視覺(jué)、多媒體技術(shù)等領(lǐng)域的應(yīng)用。具體地,本文的研究?jī)?nèi)容和主要貢獻(xiàn)分為以下四個(gè)方面：首先,根據(jù)藝術(shù)理論的相關(guān)研究,本文提出了一種基于藝術(shù)原理的中層圖像情感特征,對(duì)以圖像為中心的大眾化情感進(jìn)行預(yù)測(cè)。藝術(shù)理論由藝術(shù)元素和藝術(shù)原理組成：藝術(shù)元素是構(gòu)成藝術(shù)作品的基本元素,包括顏色、紋理等；藝術(shù)原理是用來(lái)對(duì)藝術(shù)元素進(jìn)行組織與排列的規(guī)則和工具,包括平衡、強(qiáng)調(diào)等�，F(xiàn)有的工作主要提取基于藝術(shù)元素的底層特征對(duì)圖像的情感進(jìn)行識(shí)別。這些特征容易受到組織規(guī)則的影響,并且它們與情感之間的關(guān)系很微弱。因此,藝術(shù)元素必須通過(guò)藝術(shù)原理組織排列成有意義的區(qū)域與圖像,來(lái)表達(dá)特定的語(yǔ)義與情感。本文系統(tǒng)地學(xué)習(xí)、表示并實(shí)現(xiàn)了基于藝術(shù)原理的特征,將量化后的藝術(shù)原理串聯(lián)成情感特征,用來(lái)對(duì)圖像情感進(jìn)行分類與回歸。在Abstract、 ArtPhoto三個(gè)數(shù)據(jù)集上的實(shí)驗(yàn)證明了藝術(shù)原理特征的有效性。其次,利用社交媒體上的數(shù)據(jù),本文提出了一種以用戶為中心的個(gè)性化情感預(yù)測(cè)方法,首次對(duì)圖像情感感知的主觀性進(jìn)行評(píng)價(jià)。現(xiàn)有的圖像情感數(shù)據(jù)集都是以圖像為中心的,以預(yù)測(cè)圖像情感的大眾化情感為目的,并且圖像數(shù)量很少,不能用于個(gè)性化的情感分析。本文構(gòu)造了一個(gè)基于Flickr的個(gè)性化圖像情感感知的大規(guī)模數(shù)據(jù)集,命名為Image-Emotion-Social-Net (IESN),包含100多萬(wàn)張圖像和大約8000個(gè)用戶。社交網(wǎng)絡(luò)中多種因素可以影響個(gè)性化的情感感知：視覺(jué)內(nèi)容、社交上下文、時(shí)間演變、地理位置等。本文提出了迭代多任務(wù)超圖學(xué)習(xí)方法對(duì)這些因素進(jìn)行聯(lián)合建模,并且設(shè)計(jì)了一個(gè)學(xué)習(xí)算法,實(shí)現(xiàn)自動(dòng)優(yōu)化。實(shí)驗(yàn)結(jié)果表明,綜合考慮多種因素可以有效地提高個(gè)性化情感預(yù)測(cè)的準(zhǔn)確率。再次,本文提出了一種以圖像為中心的對(duì)圖像情感的概率分布進(jìn)行預(yù)測(cè)的方法,從新的角度對(duì)圖像情感進(jìn)行建模。在Abstract以及IESN數(shù)據(jù)集上的統(tǒng)計(jì)發(fā)現(xiàn),盡管圖像情感感知呈現(xiàn)出個(gè)性化的特點(diǎn),但整體上也服從一定的分布�；�于這一觀察,本文提出了基于共享稀疏學(xué)習(xí)的方法對(duì)圖像情感的概率分布進(jìn)行預(yù)測(cè),并且使用迭代重加權(quán)最小二乘進(jìn)行優(yōu)化。對(duì)應(yīng)于離散情感和維度情感兩種表示方法,本文對(duì)圖像情感的離散概率分布和連續(xù)概率分布都進(jìn)行了處理。此外,本文介紹了多種baseline算法。實(shí)驗(yàn)結(jié)果表明,共享稀疏學(xué)習(xí)取得了最優(yōu)的性能。最后,本文實(shí)現(xiàn)了圖像情感在計(jì)算機(jī)視覺(jué)與多媒體技術(shù)領(lǐng)域的多個(gè)應(yīng)用。一個(gè)是基于多圖學(xué)習(xí)的情感圖像檢索,與傳統(tǒng)的基于內(nèi)容的圖像檢索不同,本文使用多圖學(xué)習(xí)的方法從情感的角度對(duì)圖像進(jìn)行檢索,并且在3D物體檢索上進(jìn)行了擴(kuò)充；一個(gè)是基于觀察者情感分析的視頻分類與推薦,提出了使用觀察者觀看視頻時(shí)表情的變化來(lái)對(duì)視頻進(jìn)行分析；一個(gè)是基于情感的圖像配樂(lè),為輸入圖像配置表達(dá)相似情感的音樂(lè),這可以使圖像更加生動(dòng),并且?guī)ьI(lǐng)用戶進(jìn)入圖像想要表達(dá)的世界。通過(guò)上述研究,本文對(duì)圖像情感計(jì)算的各個(gè)層面進(jìn)行了深入的探索,為圖像情感計(jì)算所面臨的關(guān)鍵問(wèn)題提供了切實(shí)有效的解決方案。結(jié)果表明：通過(guò)引入藝術(shù)學(xué)等相關(guān)學(xué)科的研究,可以提取出更具有判別力且容易理解的特征,從而提高圖像情感識(shí)別的準(zhǔn)確率；社交媒體中圖像情感的感知是個(gè)性化的,并且受到時(shí)間演變、社交上下文等多種因素的影響,綜合考慮這些因素可以顯著提高情感預(yù)測(cè)的性能；從概率分布的角度對(duì)圖像情感進(jìn)行建模,是對(duì)個(gè)性化情感與大眾化情感的折中,更符合實(shí)際情況,更具有實(shí)際意義。
[Abstract]:With the rapid development of computer science, multimedia technology and social network, the scale of multimedia content, such as image and video, is increasing exponentially. The demand for processing and understanding the multimedia content is increasing. People can only sense and understand the image, the high-level semantic layer of video, including cognition, relative to the underlying visual feature layer. The previous work on the analysis of image content is mainly focused on understanding the cognitive level of the image, that is, to describe the true content of the image, such as the object detection and recognition. However, the widespread use of digital photography and the strong demand for the emotional expression of the image make the analysis of the highest semantic layer of the image more emotional. The more urgent. The analysis of the emotional layer of the image, referred to as the image emotional calculation, is mainly to understand the emotional reaction caused by the viewer after the image. The development of the image emotion calculation is mainly restricted by the two major challenges: one is the emotional gap, that is, "the measurable signal is characteristic and human perception of the desired signal." The other is the subjectivity of human emotion perception and evaluation, that is, "the emotional perception of the same image is subjective and different from the influence of various factors such as cultural background, educational level, social context and so on." this paper studies the above problems in the image emotional calculation. Based on the theory of art, we expect to extract more perceptive emotional features, use social media data to predict the user centered personalized emotion, explore the factors that affect emotional perception in social media, model the distribution of image emotions, and predict the lure of an image in a number of observers. The distribution of emotion, the application of image emotion in computer vision, multimedia technology and other fields. Specifically, the research content and main contributions of this article are divided into four aspects: firstly, according to the related research of art theory, this paper proposes an emotional feature of middle layer image based on the principle of Art, to the image as the medium. Artistic theory consists of artistic elements and artistic principles: the elements of art constitute the basic elements of a work of art, including color, texture, etc.; the principles of art are the rules and tools used to organize and arrange the elements of art, including balance and emphasis. The existing work is mainly based on art. The underlying characteristics of the elements identify the emotions of the images. These features are easily influenced by the rules of the organization, and they have a weak relationship with the emotions. Therefore, the art elements must be organized into meaningful regions and images through the principles of art to express specific semantics and emotions. Based on the characteristics of the principle of art, the artistic principles after quantization are connected into emotional features to classify and regress the emotion of the image. The experiments on the three data sets of Abstract, ArtPhoto prove the validity of the artistic principle characteristics. Secondly, using the data on social media, this paper proposes a user centered one. The method of sexual emotion prediction is the first to evaluate the subjectivity of image emotion perception. The existing image emotional data sets are centered on the image, in order to predict the popular emotion of the image emotion, and the number of images is very small, and can not be used for personalized emotional analysis. A personalized image based on Flickr is constructed in this paper. A large dataset of emotional perception, named Image-Emotion-Social-Net (IESN), contains about 1000000 images and about 8000 users. A variety of factors in social networks can affect personalized emotional perception: visual content, social context, time evolution, geographic location, etc. This paper proposes iterative multitask hypergraph learning methods These factors are jointly modeled and a learning algorithm is designed to achieve automatic optimization. The experimental results show that a comprehensive consideration of a variety of factors can effectively improve the accuracy of personalized emotional prediction. Thirdly, this paper proposes a method for predicting the probability distribution of image emotions with image as the center, from a new angle. The image emotion is modeled. The statistics on the Abstract and the IESN data sets show that although the image emotion perception presents a personalized feature, it also obeys a certain distribution. Based on this observation, this paper proposes a method of sharing sparse learning to predict the probability distribution of image emotion, and use iterative reiteration. Weighted least squares are optimized. Corresponding to two representation methods of discrete emotion and dimension emotion, this paper deals with the discrete probability distribution and continuous probability distribution of image emotion. In addition, this paper introduces a variety of baseline algorithms. Experimental results show that shared sparse learning achieves the best performance. Finally, this paper realizes the graph. One is emotional image retrieval in the field of computer vision and multimedia technology. One is the emotional image retrieval based on multi graph learning, which is different from the traditional content based image retrieval. This paper uses the method of multi graph learning to retrieve the image from the angle of emotion, and extends the 3D physical examination cable; one is based on the view. The video classification and recommendation of the observer's emotional analysis is proposed to analyze the video using the changes of the observer to watch the video. One is the image music based on the emotion, which can express the similar emotion for the input image, which can make the image more vivid and lead the user into the world that the image wants to express. In this study, this paper makes an in-depth exploration of the various levels of image emotional computing, and provides a practical and effective solution for the key problems facing image emotion calculation. The results show that the more discriminative and easy to understand features can be extracted by introducing the related subjects of art and so on, thus improving the image. The accuracy of emotion recognition; the perceptual perception of image in social media is individualized and influenced by many factors such as time evolution, social context and so on. Considering these factors, the performance of emotion prediction can be greatly improved. The modeling of image emotion from the perspective of probability distribution is the individualized emotion and popular feeling. The compromise is more practical and practical.
【學(xué)位授予單位】：哈爾濱工業(yè)大學(xué)
【學(xué)位級(jí)別】：博士
【學(xué)位授予年份】：2016
【分類號(hào)】：TP391.41

【相似文獻(xiàn)】

相關(guān)期刊論文 前10條

1 梁柏茂;魏彪;米德伶;彭柯鑫;馬曉昕;;一種高分辨X光數(shù)字胃腸圖像系統(tǒng)的設(shè)計(jì)與實(shí)現(xiàn)[J];國(guó)外電子測(cè)量技術(shù);2007年11期

2 劉梅娟;;談地理教學(xué)中圖像系統(tǒng)的閱讀[J];教育前沿(理論版);2008年04期

3 鄭冰;展旭卿;付民;;集束光水下圖像系統(tǒng)[J];光子學(xué)報(bào);2007年02期

4 傅亨;;視頻與圖像系統(tǒng)開(kāi)發(fā)中容易出現(xiàn)的問(wèn)題與解決方法[J];信息通信;2013年08期

5 姚錦鐘，蔡清福，關(guān)立勛;超聲醫(yī)學(xué)圖像系統(tǒng)的技術(shù)現(xiàn)狀及設(shè)備水平[J];中國(guó)儀器儀表;1994年05期

6 吳啟森;基于虛擬窗口的圖像加速顯示和重畫(huà)技術(shù)[J];計(jì)算機(jī)應(yīng)用;2003年S1期

7 由春秋;;實(shí)時(shí)采集多幅圖像疊加平均去噪的研究與實(shí)現(xiàn)[J];哈爾濱師范大學(xué)自然科學(xué)學(xué)報(bào);2007年03期

8 王偉凝;蟻靜緘;賀前華;;可計(jì)算圖像美學(xué)研究進(jìn)展[J];中國(guó)圖象圖形學(xué)報(bào);2012年08期

9 向方明;朱遵義;許敬;崔業(yè)兵;;視頻/圖像系統(tǒng)開(kāi)發(fā)的關(guān)鍵問(wèn)題研究[J];現(xiàn)代電子技術(shù);2013年08期

10 ;TOP──向PostScript完全開(kāi)放[J];電子出版;1995年12期

相關(guān)會(huì)議論文 前4條

1 豐樹(shù)強(qiáng);趙瀟;張國(guó)光;;低輻射車輛檢查裝置圖像系統(tǒng)的設(shè)計(jì)與實(shí)現(xiàn)[A];第三屆全國(guó)核技術(shù)與應(yīng)用學(xué)術(shù)研討會(huì)會(huì)議資料文集[C];2012年

2 周衛(wèi)東;王翠玲;張貴倉(cāng);;一種通用醫(yī)學(xué)圖像系統(tǒng)安全模型的研究[A];中國(guó)幾何設(shè)計(jì)與計(jì)算新進(jìn)展2007——第三屆中國(guó)幾何設(shè)計(jì)與計(jì)算大會(huì)論文集[C];2007年

3 韓曉暉;麻元興;鄭穗聯(lián);;基于區(qū)域協(xié)同醫(yī)療的眼科圖像信息系統(tǒng)的研究[A];2011年浙江省眼科學(xué)術(shù)會(huì)議論文集[C];2011年

4 郭彥斌;李學(xué)先;王非;;PaxScan2520圖像系統(tǒng)在射線照相模式下的增益校正[A];2006全國(guó)荷電粒子源、粒子束學(xué)術(shù)會(huì)議論文集[C];2006年

相關(guān)重要報(bào)紙文章 前6條

1 武德鋒 李國(guó)輝 林洪文 姚作梁;圖像世界任我行[N];計(jì)算機(jī)世界;2002年

2 聶蕊;石景山 政采服務(wù)奧運(yùn)安保[N];政府采購(gòu)信息報(bào);2007年

3 陳華;打造新型醫(yī)學(xué)圖像系統(tǒng)產(chǎn)業(yè)化的鄭智捷[N];科技日?qǐng)?bào);2007年

4 王輝邋佚名;北京電子眼有了“醫(yī)生”[N];中國(guó)質(zhì)量報(bào);2007年

5 周奇;5類重點(diǎn)單位應(yīng)裝攝像頭[N];北京日?qǐng)?bào);2006年

6 ;戴爾OptiPlex SX260 體積相當(dāng)于一本字典[N];中國(guó)計(jì)算機(jī)報(bào);2002年

相關(guān)博士學(xué)位論文 前5條

1 趙思成;圖像情感感知的計(jì)算與應(yīng)用研究[D];哈爾濱工業(yè)大學(xué);2016年

2 王長(zhǎng)虎;互聯(lián)網(wǎng)環(huán)境下大規(guī)模圖像的內(nèi)容分析、檢索和自動(dòng)標(biāo)注的研究[D];中國(guó)科學(xué)技術(shù)大學(xué);2009年

3 李翎;藏傳佛教圖像研究[D];中央美術(shù)學(xué)院;2002年

4 張明宇;TDI CCD相機(jī)圖像采集與處理系統(tǒng)研究[D];中國(guó)科學(xué)院研究生院（長(zhǎng)春光學(xué)精密機(jī)械與物理研究所）;2011年

5 王江鵬;魏晉南北朝墓葬人物圖像與藝術(shù)轉(zhuǎn)型[D];西安美術(shù)學(xué)院;2012年

相關(guān)碩士學(xué)位論文 前10條

1 張璐;大陸與臺(tái)灣高中地理教材圖像系統(tǒng)比較研究[D];福建師范大學(xué);2015年

2 任曉琳;植物圖像檢索系統(tǒng)的關(guān)鍵技術(shù)研究與實(shí)現(xiàn)[D];浙江大學(xué);2015年

3 郭志威;高中地理教材圖像整合與開(kāi)發(fā)研究[D];南京師范大學(xué);2015年

4 吳龍輝;面向?yàn)?zāi)害信息提取的SAR圖像并行算法設(shè)計(jì)與實(shí)現(xiàn)[D];電子科技大學(xué);2014年

5 劉洋洋;初中生物新教材圖像系統(tǒng)特點(diǎn)及教學(xué)策略研究[D];湖南師范大學(xué);2015年

6 文潔;網(wǎng)絡(luò)環(huán)境下大學(xué)生圖像資源利用行為的調(diào)查研究[D];河南大學(xué);2015年

7 霍X;基于小波框架的Poisson數(shù)據(jù)圖像去噪算法研究[D];上海交通大學(xué);2015年

8 袁文龍;基于模型觀測(cè)器的CT圖像病灶檢測(cè)及質(zhì)量評(píng)價(jià)研究[D];東南大學(xué);2015年

9 吳彥瑾;“差圖像”的公共再現(xiàn)[D];中央美術(shù)學(xué)院;2016年

10 吳元樹(shù);利用圖像懫集裝置實(shí)現(xiàn)免疫層析定量檢測(cè)的研究[D];福州大學(xué);2014年

，

本文編號(hào)：2047834